📄 Publication: lncMAchine: A machine learning-based approach for genome-wide identification of long noncoding RNAs — Functional & Integrative Genomics, 2021
👤 Role: First author and primary developer — implemented ML pipeline and prediction framework
🎯 Impact: Enables supervised genome-wide lncRNA discovery and annotation with multiple classifiers
Tech: Python 3 • scikit-learn • BioPython • NumPy • pandas

lncMachine is designed for genome-wide identification and annotation of lncRNAs using supervised machine learning. Users can train models from coding and noncoding sequences and apply prebuilt models for prediction. lncMachine supports multiple classifiers and ensures reproducibility with reference code matching the publication.

Use cases: lncRNA annotation in new genomes • Comparative genomics • Model evaluation and classifier benchmarking

• Python 3 or newer
• scikit-learn version 0.22
• BioPython
• Numpy
• pandas

Clone the repository and ensure dependencies are installed:

git clone https://github.com/hbusra/lncMAchine.git
cd lncMAchine
pip install -r requirements.txt

Note: Prebuilt models require scikit-learn 0.22 for compatibility.

Train a Random Forest prediction model from coding and noncoding FASTA files:

python3 lncMachine.py -c coding.fasta -n noncoding.fasta --train 

Train prediction models with nine ML algorithms:

python3 lncMachine.py -c coding.fasta -n noncoding.fasta --train --all

Train from a CSV feature file:

python3 lncMachine.py -i features.csv --train

Predict coding probability from a FASTA file using a prebuilt model:

python3 lncMachine.py -c test.fasta --model prebuilt_model.sav -o test_predictions.csv

Cagirici et al., “lncMachine: a machine learning-based approach for genome-wide identification of long noncoding RNAs” Functional & Integrative Genomics, 2021